[Design of Messaging on JBoss (Messaging/JBoss)] - Re: AIO/DirectIO Integration (Repost)
by clebert.suconic@jboss.com
The test is a simple copy of testMultipleTransactionsDifferentIDs that started to fail when I was using AIO (as it changed the positioning of the records a little bit).
If the test is illegl, a the delete should throw an exception, but the test is failing just because of the position of the commit. In fact I managed to change the test in another way and still fails:
| addTx(1, 1, 2, 3, 4, 5, 6);
| updateTx(1, 1, 3, 5);
| commit(1);
| deleteTx(2, 1, 2, 3, 4, 5, 6);
| commit(2);
|
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4146807#4146807
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4146807
16 years, 4 months
[Design of JBoss Web Services] - UsernameToken authentication and authorization for POJO endp
by darran.lofthouse@jboss.com
I am thinking about having a look at this issue and just wanted to bring up some ideas here. The reason I am looking at this is because although there is a solution based on using EJB endpoints there is still a consistent demand for this capability for POJO endpoints.
We currently have the following unscheduled issue: -
http://jira.jboss.org/jira/browse/JBWS-1999
I have seen the contributed code but this does not integrate with our current WS-Security handlers so I am proposing a more integrated solution.
My idea would be to re-open the following issue to allow the UsernameToken to be set as a requirement on the incoming message: -
http://jira.jboss.org/jira/browse/JBWS-1136
The configuration should have an attribute 'authenicate=true', if set we can make use of the programatic web authentication available from JBoss 4.2.0.GA: -
http://wiki.jboss.org/wiki/WebAuthentication
In addition to this the configuration could then contain a set of the allowed roles to call the endpoint and if this is set after the authentication we could use isCallerInRole to verify if the user is in the allowed role.
The use of the WebAuthentication above does mean that we can mainly use the standard servlet APIs after the authentication and this change would be achieved with a small amount of additional configuration, as we have authenticated then this will still be propagated to the calls to any subsequent EJBs.
I will need to consider the implications of this if a user enables it for an EJB endpoint as it does depend on the web app having a security domain but the primary purpose of this change is for POJO endpoints and not EJB endpoints.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4146806#4146806
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4146806
16 years, 4 months
[Design of JBoss ESB] - Re: Publish ESB services to web services
by Kevin.Conner@jboss.com
"jim.ma" wrote : The mainly intention for it is we can make the Request/Response class as a subclass of esb Message.
But why is this a requirement? What is wrong with using POJOs and populating a created message?
"jim.ma" wrote : That's I need to consider . I do not know if it can do this if we map the request object filed name to name location and the filed value to ESB part value. For example :
| org.jboss.esb.Request extends XMLMessageBase {
| public String foo;
| public byte[] bar;
| }
To fit in with the current code it would be better if it was handled as follows.
payloadProxy = new MessagePayloadProxy(config)
and then something like the following
| Message message = MessageFactory.getInstance().getMessage();
| payloadProxy.setPayload(message, <incoming request>);
| ... evaluate pipeline ...
| Object response = payloadProxy.getPayload(responseMessage);
|
The payload proxy handles the details of previous message variants and also allows the message locations to be specified.
One thing that would be required would be a change in the way the pipeline works. At present the response/fault is handled asynchronously within the pipeline processing and we would need to refactor this code to allow for request/response processing.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4146775#4146775
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4146775
16 years, 4 months
[Design of JBoss ESB] - Re: Publish ESB services to web services
by jim.ma
anonymous wrote : What is the reason for inheriting from a base class? Would it not be easier if these were POJOs?
The mainly intention for it is we can make the Request/Response class as a subclass of esb Message.
If we create the Request object , we can directly use this object to invoke ESB message. After we exposed esb service as a web service , we can create this request object and use this class to marshal soap message like the jaxws client does.
anonymous wrote : think it would be better to use a named location within the message body for both parts and allow the names to be configurable. This would then be consistent with the way we are handling similar tasks within the current codebase, for example the integration with jBPM/drools maps these named locations into/out of their variable scopes.
That's I need to consider . I do not know if it can do this if we map the request object filed name to name location and the filed value to ESB part value. For example :
org.jboss.esb.Request extends XMLMessageBase {
public String foo;
public byte[] bar;
}
In XMLMessageBase class , we can ship the foo value and bar value into message body : getBody().add(field.getName(), filedValue);
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4146759#4146759
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4146759
16 years, 4 months
[Design of JBoss ESB] - Re: Publish ESB services to web services
by Kevin.Conner@jboss.com
"jim.ma" wrote : org.jboss.esb.Request , Response and Fault class can be a java bean and inherits from a mew class XMLMessageBase .
What is the reason for inheriting from a base class? Would it not be easier if these were POJOs?
"jim.ma" wrote : In XMLMessageBase, we can put the request part value in Esb message body . Also by invoking the method in XMLMessageBase, we can retrieve the response part value from esb message body .
I think it would be better to use a named location within the message body for both parts and allow the names to be configurable. This would then be consistent with the way we are handling similar tasks within the current codebase, for example the integration with jBPM/drools maps these named locations into/out of their variable scopes.
"jim.ma" wrote : By reflecting the Request class, we can generate schema using apache XMLSchema even there is no annotation in it.
This sounds like a good idea.
Kev
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4146753#4146753
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4146753
16 years, 4 months